Good point at the end, skip the rest if you aren't interested in the argument.
goran.hultgren@bluefish.se wrote: > Looking at the code, I see that "mail to list" sends the message > to "squeak@cs.uiuc.edu". I note that the originating address for Eh. Where did you look? I just looked in my 3.2g image and I ended up in: ChangeSet>>buildMessageForMailOutWithUser: ...and as far back as 9/27/2001 (#4385) it seems to be correct. Mind though I might be looking in the wrong place. The address is in my image: squeak-dev@lists.squeakfoundation.org What update-level is your image? Squeak 3.0.1 of 2001.02.04, update 3552.
Sure could! In that case I guess that Squeak fired it away alright but you didn't get to see the bounce(?) It was simpler and far more embarassing than that. See bottom.
> Ear sick horse. Mail sent from Netscape worked fine. What does "Ear sick horse" mean? I am not native english speaking so I didn't get that. "Ear sick horse" = "Yes of course".
> - they are not necessary, > - they are not supported by all mail user agents, > - they are notorious for insecurity. Just curious, how is uuencoding better than attachments then? It's not supported in all email clients AFAIK. The key point is that it doesn't _have_ to be.
So you want us to change this?
Obviously not. Why should I want to stop anybody doing anything that works for them? What I was hoping for was a more generous attitude. For example, for someone who understands Celeste, it should be a simple task to write a filter that decodes a uuencoded message and fires it out again as a Celeste attachment. Less than half the energy that has been spent in stepping on me could have given us the best of both worlds.
> I am of course perfectly willing to use whatever method *works*. Well, you could have fooled me. But you say you read the bit where I said I had made repeated attempts to try the official way, with no success. How could such statements possibly fool you into thinking I was not willing to use a working method?
OK, HERE's THE GOOD POINT.
There are two ways of using mail services. I suppose you might call them the "institutional user" and "home user" ways.
I've been using E-mail systems since 1979, always as an "institutional user". In this setting, which is the original RFC 821 (SMPT) setting, when you tell your mail user agent "send this message", it may send it immediately, or it may put it in a "spooling" area and send it later, but not _much_ later, typically at most an hour or two. As far as you are concerned, the message is _sent_ immediately. Maybe it's resting at a local "post office" (the spooling area), but it's already on it's way right now.
In the "home user" setting, your mail is held on some other machine. You have to explicitly "pull down" your mail (like a trip to the post office to check a post box) and when you send mail, you are just adding it to a "mail bag" which will NOT be delivered to the "post office" (this time your ISP) until you send it. So "send a message" in this environment seems to mean "I have finished this letter, put it in the mail bag" and is a very different operation from "send this mail BAG to the post office".
From my point of view, what happened is that Celeste lied to me.
Remember, I have Celeste on a Sparc and a Mac, but neither of them is my mail machine. On the rare occasions when I send mail from either of them using Netscape, it either didn't work (Sparc) or sent the message _immediately_ (Mac). So when I saw a great big button labelled "send message", I not unreasonably expected it to SEND THE MESSAGE. This was the lie. That button does NOT mean "send the message", it means "add the message to the .tosend category". The message will _never_ be sent if you don't then go to another window (which, don't forget, is _not_ the change sorter you started it, _nor_ the message editing window you've just finished with, but a Celeste control panel you may never have heard of before) and select ".tosend" (another thing you have never heard of, is not explained anywhere, and which you have no reason to suspect may be of interest) and then do the right thing, nothing will happen.
If you _do_ discover that '.tosend' exists, open it up, select the message you just asked to be sent, and check the menu, you will find nothing there about sending the message. You will conclude, "Oh, this works like some other mailers, it builds up a queue and then sends it later AUTOMATICALLY." So you wait, and wait, and wait, and nothing happens.
The way I finally figured this out is that someone got me to check whether mail sent to the STMP server using Telnet worked, and it did. So I wondered "has Celeste got the right server address", poked around a bit, tried 'Celeste smptServer' and found that it had never been initialised. So it was clear that Celeste had never tried to send the mail. (After my earlier unsuccessful attempts, I had deleted Celeste's files, assuming there was nothing interesting them. A mistake on my part.) I then tried _everything_ in the control panel, and eventually discovered 'send queued mail' in the menu on '.tosend' (NOT on the things that '.tosend' points to).
Apparently Celeste users are happy with the way it works. I have not the slightest wish that they should be made to do anything different. However, in my environment, where the STMP server is on the same network as the machine running Squeak, there is absolutely no advantage in queueing the mail. (Except of course that you have one more chance to avoid sending a bad-mannered message.) There is, in contrast, a big advantage in sending mail right away. That is that Macintoshes have been known to crash. (As noted before, while I was trying to find out what was going on, Squeak (+ perhaps Acrobat Reader) crashed my Mac twice.) The sooner a message is sent, the less chance a crash will get in the way.
It must be admitted that if you let the mouse hover over 'Send message' for long enough, eventually you will get a help balloon that says in small letters "add this to the queue of messages to be sent". But when you see "Send message", you don't suspect that you don't know what it means, so you just click away and DON'T wait for a help balloon.
I propose the following change to Celeste. For all I know, it may already have been made in Squeak 3.2, but if not, I hope someone will do it.
A "Mister Postman" window has two big buttons at the top, "send message" and "add attachment". Change this to three: "send message later", "send message now", and "add attachment". "send message now" should open up an SMPT socket and send the message right away.
On Tue, 30 Apr 2002, Richard A. O'Keefe wrote:
So when I saw a great big button labelled "send message", I not unreasonably expected it to SEND THE MESSAGE. This was the lie. That button does NOT mean "send the message", it means "add the message to the .tosend category". [...] Apparently Celeste users are happy with the way it works.
Indeed, this is what Celeste users expect. What happened was that you were classified as "Celeste user" because you had, as you admitted, tried to use Celeste before.
If you had *never* before used Celeste, then "send message" would have sent the message out immediately. I just checked, it works.
I propose the following change to Celeste. For all I know, it may already have been made in Squeak 3.2, but if not, I hope someone will do it.
A "Mister Postman" window has two big buttons at the top, "send message" and "add attachment". Change this to three: "send message later", "send message now", and "add attachment". "send message now" should open up an SMPT socket and send the message right away.
I second that, or a similar arrangement. Maybe the "send" button could just be labeled depending upon what it actually will do, that is, if you are classified as Celeste user or not. For example, "move to .tosend" vs. "send now".
-- Bert
Bert Freudenberg wrote:
On Tue, 30 Apr 2002, Richard A. O'Keefe wrote:
So when I saw a great big button labelled "send message", I not unreasonably expected it to SEND THE MESSAGE. This was the lie. That button does NOT mean "send the message", it means "add the message to the .tosend category". [...] Apparently Celeste users are happy with the way it works.
Indeed, this is what Celeste users expect. What happened was that you were classified as "Celeste user" because you had, as you admitted, tried to use Celeste before.
If you had *never* before used Celeste, then "send message" would have sent the message out immediately. I just checked, it works.
The same UI widget labeled the same way can do either of two very different things depending on something that may have happened in the distant past? Yuck!
I propose the following change to Celeste. For all I know, it may already have been made in Squeak 3.2, but if not, I hope someone will do it.
A "Mister Postman" window has two big buttons at the top, "send message" and "add attachment". Change this to three: "send message later", "send message now", and "add attachment". "send message now" should open up an SMPT socket and send the message right away.
I second that, or a similar arrangement. Maybe the "send" button could just be labeled depending upon what it actually will do, that is, if you are classified as Celeste user or not. For example, "move to .tosend" vs. "send now".
That would probably be an easier fix, but I'd prefer the extra button. (For non-Celeste users, I suppose the queue button might not appear.)
-Jesse
At 9:45 AM +0200 4/30/02, Bert Freudenberg wrote:
A "Mister Postman" window has two big buttons at the top, "send message" and "add attachment". Change this to three: "send message later", "send message now", and "add attachment". "send message now" should open up an SMPT socket and send the message right away.
I second that, or a similar arrangement. Maybe the "send" button could just be labeled depending upon what it actually will do, that is, if you are classified as Celeste user or not. For example, "move to .tosend" vs. "send now".
Eudora's designers made a similar choice. The button is labeled either "Send" or "Queue" depending on how you have your preferences set. I tend to think that's a reasonable design.
-Martin
On Tue, 30 Apr 2002, Richard A. O'Keefe wrote:
Good point at the end, skip the rest if you aren't interested in the argument.
[snip]
OK, HERE's THE GOOD POINT.
There are two ways of using mail services. I suppose you might call them the "institutional user" and "home user" ways.
Furthermore, there are at least two sorts of submitters...those who use Celeste and those who don't want to. The latter may include those who use an alternative client or just don't want to use *any* client in Squeak. And they don't have to. Squeak can mail stuff directly and immediately.
I've been using E-mail systems since 1979, always as an "institutional user". In this setting, which is the original RFC 821 (SMPT) setting, when you tell your mail user agent "send this message", it may send it immediately, or it may put it in a "spooling" area and send it later, but not _much_ later, typically at most an hour or two. As far as you are concerned, the message is _sent_ immediately. Maybe it's resting at a local "post office" (the spooling area), but it's already on it's way right now.
I submitted a goodie a ways back (I think) that wrote right to SMTPSocket instead of queuing in Celeste. Thus, if you dumped Celeste from your image you could use mail to list.
[snip]
Apparently Celeste users are happy with the way it works. I have not the slightest wish that they should be made to do anything different. However, in my environment, where the STMP server is on the same network as the machine running Squeak, there is absolutely no advantage in queueing the mail. (Except of course that you have one more chance to avoid sending a bad-mannered message.) There is, in contrast, a big advantage in sending mail right away.
Definitely. And there are other advantages to using Celeste. I often have several images open at a time, but I only want one to be my "celeste image", yet I'd still like to submit from all these images without having to throw change sets around (and, as far as I know, there's no mail to list from the filelist). There's no clean way to bundle muliple discrete changesets (and, oh, documentation files), etc. etc. etc.
Plus, there are other places one might want to submit (to each other, to the SuperSwiki, to SCAN, to the Swiki, etc.). And perhaps more "types" of things to submit (i.e., test cases, comments, docs, etc.)
Anyhoo, what I'm suggesting is that taking a step back and looking at the whole framework might be a good idea. If we do it right, then upgrading, oh, Bert's tools shouldn't be bad at all.
[snip]
I propose the following change to Celeste. For all I know, it may already have been made in Squeak 3.2, but if not, I hope someone will do it.
A "Mister Postman" window has two big buttons at the top, "send message" and "add attachment". Change this to three: "send message later", "send message now", and "add attachment". "send message now" should open up an SMPT socket and send the message right away.
Ah, here's my [ENH] (sorry, not a goodie):
http://swiki.gsug.org:8080/sqfixes/1528.html
Yay sqfixes!
It may not entirely work as is as it prolly uses older preference structure.
Note it didn't do *quite* what you asked, as it avoids Celeste altogether.
Cheers, Bijan Parsia.
Bijan Parsia wrote:
[snip]
I propose the following change to Celeste. For all I know, it may already have been made in Squeak 3.2, but if not, I hope someone will do it.
A "Mister Postman" window has two big buttons at the top, "send message" and "add attachment". Change this to three: "send message later", "send message now", and "add attachment". "send message now" should open up an SMTP socket and send the message right away.
Ah, here's my [ENH] (sorry, not a goodie):
http://swiki.gsug.org:8080/sqfixes/1528.html
Yay sqfixes!
It may not entirely work as is as it prolly uses older preference structure.
Note it didn't do *quite* what you asked, as it avoids Celeste altogether.
Actually, this changeset has already been incorporated into the main release (sometime during 3.1), with some minor changes.
If you don't have anything with Celeste configured, and you try to "mail to list", it prompts you to enter a temporary email address and SMTP server, so you don't have to mess with Celeste if you don't want to.
The sender's full name feature was originally incorporated too, but then it was removed because it caused some sort of problem with some SMTP servers, I'm not sure why.
- Doug Way dway@riskmetrics.com
On Tue, 30 Apr 2002, Doug Way wrote:
Bijan Parsia wrote:
[snip]
Ah, here's my [ENH] (sorry, not a goodie):
http://swiki.gsug.org:8080/sqfixes/1528.html
Yay sqfixes!
It may not entirely work as is as it prolly uses older preference structure.
Note it didn't do *quite* what you asked, as it avoids Celeste altogether.
Actually, this changeset has already been incorporated into the main release (sometime during 3.1), with some minor changes.
D'oh! Wow, wipe the egg off *my* face.
If you don't have anything with Celeste configured, and you try to "mail to list", it prompts you to enter a temporary email address and SMTP server, so you don't have to mess with Celeste if you don't want to.
Ah. This is what was meant by mysterious change! Yes, I was finding this behavior *quite* perplexing!
Hmm. Note that in my [ENH] this was an explicit preference which, imho, is a little nicer/safer.
Well, anyhoo! I'm willing to improve this a bit along the lines people have been suggesting, and maybe add some warnings/documentation. Then I can harvest it :)
The sender's full name feature was originally incorporated too, but then it was removed because it caused some sort of problem with some SMTP servers, I'm not sure why.
Ok.
Cheers, Bijan Parsia.
squeak-dev@lists.squeakfoundation.org